10 pvecm - Proxmox VE Cluster Manager
15 include::pvecm.1-synopsis.adoc[]
27 The {PVE} cluster manager `pvecm` is a tool to create a group of
28 physical servers. Such a group is called a *cluster*. We use the
29 http://www.corosync.org[Corosync Cluster Engine] for reliable group
30 communication, and such clusters can consist of up to 32 physical nodes
31 (probably more, dependent on network latency).
33 `pvecm` can be used to create a new cluster, join nodes to a cluster,
34 leave the cluster, get status information and do various other cluster
35 related tasks. The **P**rox**m**o**x** **C**luster **F**ile **S**ystem (``pmxcfs'')
36 is used to transparently distribute the cluster configuration to all cluster
39 Grouping nodes into a cluster has the following advantages:
41 * Centralized, web based management
43 * Multi-master clusters: each node can do all management task
45 * `pmxcfs`: database-driven file system for storing configuration files,
46 replicated in real-time on all nodes using `corosync`.
48 * Easy migration of virtual machines and containers between physical
53 * Cluster-wide services like firewall and HA
59 * All nodes must be in the same network as `corosync` uses IP Multicast
60 to communicate between nodes (also see
61 http://www.corosync.org[Corosync Cluster Engine]). Corosync uses UDP
62 ports 5404 and 5405 for cluster communication.
64 NOTE: Some switches do not support IP multicast by default and must be
65 manually enabled first.
67 * Date and time have to be synchronized.
69 * SSH tunnel on TCP port 22 between nodes is used.
71 * If you are interested in High Availability, you need to have at
72 least three nodes for reliable quorum. All nodes should have the
75 * We recommend a dedicated NIC for the cluster traffic, especially if
76 you use shared storage.
78 * Root password of a cluster node is required for adding nodes.
80 NOTE: It is not possible to mix Proxmox VE 3.x and earlier with
81 Proxmox VE 4.0 cluster nodes.
87 First, install {PVE} on all nodes. Make sure that each node is
88 installed with the final hostname and IP configuration. Changing the
89 hostname and IP is not possible after cluster creation.
91 Currently the cluster creation can either be done on the console (login via
92 `ssh`) or the API, which we have a GUI implementation for (__Datacenter ->
95 [[pvecm_create_cluster]]
99 Login via `ssh` to the first {pve} node. Use a unique name for your cluster.
100 This name cannot be changed later. The cluster name follows the same rules as node names.
102 hp1# pvecm create YOUR-CLUSTER-NAME
104 CAUTION: The cluster name is used to compute the default multicast
105 address. Please use unique cluster names if you run more than one
106 cluster inside your network.
108 To check the state of your cluster use:
112 Multiple Clusters In Same Network
113 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
115 It is possible to create multiple clusters in the same physical or logical
116 network. Each cluster must have a unique name, which is used to generate the
117 cluster's multicast group address. As long as no duplicate cluster names are
118 configured in one network segment, the different clusters won't interfere with
121 If multiple clusters operate in a single network it may be beneficial to setup
122 an IGMP querier and enable IGMP Snooping in said network. This may reduce the
123 load of the network significantly because multicast packets are only delivered
124 to endpoints of the respective member nodes.
127 [[pvecm_join_node_to_cluster]]
128 Adding Nodes to the Cluster
129 ---------------------------
131 Login via `ssh` to the node you want to add.
133 hp2# pvecm add IP-ADDRESS-CLUSTER
135 For `IP-ADDRESS-CLUSTER` use the IP from an existing cluster node.
137 CAUTION: A new node cannot hold any VMs, because you would get
138 conflicts about identical VM IDs. Also, all existing configuration in
139 `/etc/pve` is overwritten when you join a new node to the cluster. To
140 workaround, use `vzdump` to backup and restore to a different VMID after
141 adding the node to the cluster.
143 To check the state of cluster:
147 .Cluster status after adding 4 nodes
152 Date: Mon Apr 20 12:30:13 2015
153 Quorum provider: corosync_votequorum
159 Votequorum information
160 ~~~~~~~~~~~~~~~~~~~~~~
167 Membership information
168 ~~~~~~~~~~~~~~~~~~~~~~
170 0x00000001 1 192.168.15.91
171 0x00000002 1 192.168.15.92 (local)
172 0x00000003 1 192.168.15.93
173 0x00000004 1 192.168.15.94
176 If you only want the list of all nodes use:
180 .List nodes in a cluster
184 Membership information
185 ~~~~~~~~~~~~~~~~~~~~~~
193 [[adding-nodes-with-separated-cluster-network]]
194 Adding Nodes With Separated Cluster Network
195 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
197 When adding a node to a cluster with a separated cluster network you need to
198 use the 'ringX_addr' parameters to set the nodes address on those networks:
202 pvecm add IP-ADDRESS-CLUSTER -ring0_addr IP-ADDRESS-RING0
205 If you want to use the Redundant Ring Protocol you will also want to pass the
206 'ring1_addr' parameter.
209 Remove a Cluster Node
210 ---------------------
212 CAUTION: Read carefully the procedure before proceeding, as it could
213 not be what you want or need.
215 Move all virtual machines from the node. Make sure you have no local
216 data or backups you want to keep, or save them accordingly.
217 In the following example we will remove the node hp4 from the cluster.
219 Log in to a *different* cluster node (not hp4), and issue a `pvecm nodes`
220 command to identify the node ID to remove:
225 Membership information
226 ~~~~~~~~~~~~~~~~~~~~~~
235 At this point you must power off hp4 and
236 make sure that it will not power on again (in the network) as it
239 IMPORTANT: As said above, it is critical to power off the node
240 *before* removal, and make sure that it will *never* power on again
241 (in the existing cluster network) as it is.
242 If you power on the node as it is, your cluster will be screwed up and
243 it could be difficult to restore a clean cluster state.
245 After powering off the node hp4, we can safely remove it from the cluster.
247 hp1# pvecm delnode hp4
249 If the operation succeeds no output is returned, just check the node
250 list again with `pvecm nodes` or `pvecm status`. You should see
258 Date: Mon Apr 20 12:44:28 2015
259 Quorum provider: corosync_votequorum
265 Votequorum information
266 ~~~~~~~~~~~~~~~~~~~~~~
273 Membership information
274 ~~~~~~~~~~~~~~~~~~~~~~
276 0x00000001 1 192.168.15.90 (local)
277 0x00000002 1 192.168.15.91
278 0x00000003 1 192.168.15.92
281 If, for whatever reason, you want that this server joins the same
282 cluster again, you have to
284 * reinstall {pve} on it from scratch
286 * then join it, as explained in the previous section.
288 [[pvecm_separate_node_without_reinstall]]
289 Separate A Node Without Reinstalling
290 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
292 CAUTION: This is *not* the recommended method, proceed with caution. Use the
293 above mentioned method if you're unsure.
295 You can also separate a node from a cluster without reinstalling it from
296 scratch. But after removing the node from the cluster it will still have
297 access to the shared storages! This must be resolved before you start removing
298 the node from the cluster. A {pve} cluster cannot share the exact same
299 storage with another cluster, as storage locking doesn't work over cluster
300 boundary. Further, it may also lead to VMID conflicts.
302 Its suggested that you create a new storage where only the node which you want
303 to separate has access. This can be an new export on your NFS or a new Ceph
304 pool, to name a few examples. Its just important that the exact same storage
305 does not gets accessed by multiple clusters. After setting this storage up move
306 all data from the node and its VMs to it. Then you are ready to separate the
307 node from the cluster.
309 WARNING: Ensure all shared resources are cleanly separated! You will run into
310 conflicts and problems else.
312 First stop the corosync and the pve-cluster services on the node:
315 systemctl stop pve-cluster
316 systemctl stop corosync
319 Start the cluster filesystem again in local mode:
325 Delete the corosync configuration files:
328 rm /etc/pve/corosync.conf
332 You can now start the filesystem again as normal service:
336 systemctl start pve-cluster
339 The node is now separated from the cluster. You can deleted it from a remaining
340 node of the cluster with:
343 pvecm delnode oldnode
346 If the command failed, because the remaining node in the cluster lost quorum
347 when the now separate node exited, you may set the expected votes to 1 as a workaround:
353 And the repeat the 'pvecm delnode' command.
355 Now switch back to the separated node, here delete all remaining files left
356 from the old cluster. This ensures that the node can be added to another
357 cluster again without problems.
361 rm /var/lib/corosync/*
364 As the configuration files from the other nodes are still in the cluster
365 filesystem you may want to clean those up too. Remove simply the whole
366 directory recursive from '/etc/pve/nodes/NODENAME', but check three times that
367 you used the correct one before deleting it.
369 CAUTION: The nodes SSH keys are still in the 'authorized_key' file, this means
370 the nodes can still connect to each other with public key authentication. This
371 should be fixed by removing the respective keys from the
372 '/etc/pve/priv/authorized_keys' file.
377 {pve} use a quorum-based technique to provide a consistent state among
380 [quote, from Wikipedia, Quorum (distributed computing)]
382 A quorum is the minimum number of votes that a distributed transaction
383 has to obtain in order to be allowed to perform an operation in a
387 In case of network partitioning, state changes requires that a
388 majority of nodes are online. The cluster switches to read-only mode
391 NOTE: {pve} assigns a single vote to each node by default.
396 The cluster network is the core of a cluster. All messages sent over it have to
397 be delivered reliable to all nodes in their respective order. In {pve} this
398 part is done by corosync, an implementation of a high performance low overhead
399 high availability development toolkit. It serves our decentralized
400 configuration file system (`pmxcfs`).
402 [[cluster-network-requirements]]
405 This needs a reliable network with latencies under 2 milliseconds (LAN
406 performance) to work properly. While corosync can also use unicast for
407 communication between nodes its **highly recommended** to have a multicast
408 capable network. The network should not be used heavily by other members,
409 ideally corosync runs on its own network.
410 *never* share it with network where storage communicates too.
412 Before setting up a cluster it is good practice to check if the network is fit
415 * Ensure that all nodes are in the same subnet. This must only be true for the
416 network interfaces used for cluster communication (corosync).
418 * Ensure all nodes can reach each other over those interfaces, using `ping` is
419 enough for a basic test.
421 * Ensure that multicast works in general and a high package rates. This can be
422 done with the `omping` tool. The final "%loss" number should be < 1%.
426 omping -c 10000 -i 0.001 -F -q NODE1-IP NODE2-IP ...
429 * Ensure that multicast communication works over an extended period of time.
430 This uncovers problems where IGMP snooping is activated on the network but
431 no multicast querier is active. This test has a duration of around 10
436 omping -c 600 -i 1 -q NODE1-IP NODE2-IP ...
439 Your network is not ready for clustering if any of these test fails. Recheck
440 your network configuration. Especially switches are notorious for having
441 multicast disabled by default or IGMP snooping enabled with no IGMP querier
444 In smaller cluster its also an option to use unicast if you really cannot get
447 Separate Cluster Network
448 ~~~~~~~~~~~~~~~~~~~~~~~~
450 When creating a cluster without any parameters the cluster network is generally
451 shared with the Web UI and the VMs and its traffic. Depending on your setup
452 even storage traffic may get sent over the same network. Its recommended to
453 change that, as corosync is a time critical real time application.
455 Setting Up A New Network
456 ^^^^^^^^^^^^^^^^^^^^^^^^
458 First you have to setup a new network interface. It should be on a physical
459 separate network. Ensure that your network fulfills the
460 <<cluster-network-requirements,cluster network requirements>>.
462 Separate On Cluster Creation
463 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
465 This is possible through the 'ring0_addr' and 'bindnet0_addr' parameter of
466 the 'pvecm create' command used for creating a new cluster.
468 If you have setup an additional NIC with a static address on 10.10.10.1/25
469 and want to send and receive all cluster communication over this interface
474 pvecm create test --ring0_addr 10.10.10.1 --bindnet0_addr 10.10.10.0
477 To check if everything is working properly execute:
480 systemctl status corosync
483 Afterwards, proceed as descripted in the section to
484 <<adding-nodes-with-separated-cluster-network,add nodes with a separated cluster network>>.
486 [[separate-cluster-net-after-creation]]
487 Separate After Cluster Creation
488 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
490 You can do this also if you have already created a cluster and want to switch
491 its communication to another network, without rebuilding the whole cluster.
492 This change may lead to short durations of quorum loss in the cluster, as nodes
493 have to restart corosync and come up one after the other on the new network.
495 Check how to <<edit-corosync-conf,edit the corosync.conf file>> first.
496 The open it and you should see a file similar to:
530 provider: corosync_votequorum
534 cluster_name: thomas-testcluster
540 bindnetaddr: 192.168.30.50
547 The first you want to do is add the 'name' properties in the node entries if
548 you do not see them already. Those *must* match the node name.
550 Then replace the address from the 'ring0_addr' properties with the new
551 addresses. You may use plain IP addresses or also hostnames here. If you use
552 hostnames ensure that they are resolvable from all nodes.
554 In my example I want to switch my cluster communication to the 10.10.10.1/25
555 network. So I replace all 'ring0_addr' respectively. I also set the bindnetaddr
556 in the totem section of the config to an address of the new network. It can be
557 any address from the subnet configured on the new network interface.
559 After you increased the 'config_version' property the new configuration file
575 ring0_addr: 10.10.10.2
582 ring0_addr: 10.10.10.3
589 ring0_addr: 10.10.10.1
595 provider: corosync_votequorum
599 cluster_name: thomas-testcluster
605 bindnetaddr: 10.10.10.1
612 Now after a final check whether all changed information is correct we save it
613 and see again the <<edit-corosync-conf,edit corosync.conf file>> section to
614 learn how to bring it in effect.
616 As our change cannot be enforced live from corosync we have to do an restart.
618 On a single node execute:
621 systemctl restart corosync
624 Now check if everything is fine:
628 systemctl status corosync
631 If corosync runs again correct restart corosync also on all other nodes.
632 They will then join the cluster membership one by one on the new network.
635 Redundant Ring Protocol
636 ~~~~~~~~~~~~~~~~~~~~~~~
637 To avoid a single point of failure you should implement counter measurements.
638 This can be on the hardware and operating system level through network bonding.
640 Corosync itself offers also a possibility to add redundancy through the so
641 called 'Redundant Ring Protocol'. This protocol allows running a second totem
642 ring on another network, this network should be physically separated from the
643 other rings network to actually increase availability.
645 RRP On Cluster Creation
646 ~~~~~~~~~~~~~~~~~~~~~~~
648 The 'pvecm create' command provides the additional parameters 'bindnetX_addr',
649 'ringX_addr' and 'rrp_mode', can be used for RRP configuration.
651 NOTE: See the <<corosync-conf-glossary,glossary>> if you do not know what each parameter means.
653 So if you have two networks, one on the 10.10.10.1/24 and the other on the
654 10.10.20.1/24 subnet you would execute:
658 pvecm create CLUSTERNAME -bindnet0_addr 10.10.10.1 -ring0_addr 10.10.10.1 \
659 -bindnet1_addr 10.10.20.1 -ring1_addr 10.10.20.1
662 RRP On Existing Clusters
663 ~~~~~~~~~~~~~~~~~~~~~~~~
665 You will take similar steps as described in
666 <<separate-cluster-net-after-creation,separating the cluster network>> to
667 enable RRP on an already running cluster. The single difference is, that you
668 will add `ring1` and use it instead of `ring0`.
670 First add a new `interface` subsection in the `totem` section, set its
671 `ringnumber` property to `1`. Set the interfaces `bindnetaddr` property to an
672 address of the subnet you have configured for your new ring.
673 Further set the `rrp_mode` to `passive`, this is the only stable mode.
675 Then add to each node entry in the `nodelist` section its new `ring1_addr`
676 property with the nodes additional ring address.
678 So if you have two networks, one on the 10.10.10.1/24 and the other on the
679 10.10.20.1/24 subnet, the final configuration file should look like:
690 bindnetaddr: 10.10.10.1
694 bindnetaddr: 10.10.20.1
704 ring0_addr: 10.10.10.1
705 ring1_addr: 10.10.20.1
712 ring0_addr: 10.10.10.2
713 ring1_addr: 10.10.20.2
716 [...] # other cluster nodes here
719 [...] # other remaining config sections here
723 Bring it in effect like described in the
724 <<edit-corosync-conf,edit the corosync.conf file>> section.
726 This is a change which cannot take live in effect and needs at least a restart
727 of corosync. Recommended is a restart of the whole cluster.
729 If you cannot reboot the whole cluster ensure no High Availability services are
730 configured and the stop the corosync service on all nodes. After corosync is
731 stopped on all nodes start it one after the other again.
733 Corosync Configuration
734 ----------------------
736 The `/etc/pve/corosync.conf` file plays a central role in {pve} cluster. It
737 controls the cluster member ship and its network.
738 For reading more about it check the corosync.conf man page:
744 For node membership you should always use the `pvecm` tool provided by {pve}.
745 You may have to edit the configuration file manually for other changes.
746 Here are a few best practice tips for doing this.
748 [[edit-corosync-conf]]
752 Editing the corosync.conf file can be not always straight forward. There are
753 two on each cluster, one in `/etc/pve/corosync.conf` and the other in
754 `/etc/corosync/corosync.conf`. Editing the one in our cluster file system will
755 propagate the changes to the local one, but not vice versa.
757 The configuration will get updated automatically as soon as the file changes.
758 This means changes which can be integrated in a running corosync will take
759 instantly effect. So you should always make a copy and edit that instead, to
760 avoid triggering some unwanted changes by an in between safe.
764 cp /etc/pve/corosync.conf /etc/pve/corosync.conf.new
767 Then open the Config file with your favorite editor, `nano` and `vim.tiny` are
768 preinstalled on {pve} for example.
770 NOTE: Always increment the 'config_version' number on configuration changes,
771 omitting this can lead to problems.
773 After making the necessary changes create another copy of the current working
774 configuration file. This serves as a backup if the new configuration fails to
775 apply or makes problems in other ways.
779 cp /etc/pve/corosync.conf /etc/pve/corosync.conf.bak
782 Then move the new configuration file over the old one:
785 mv /etc/pve/corosync.conf.new /etc/pve/corosync.conf
788 You may check with the commands
791 systemctl status corosync
792 journalctl -b -u corosync
795 If the change could applied automatically. If not you may have to restart the
796 corosync service via:
799 systemctl restart corosync
802 On errors check the troubleshooting section below.
807 Issue: 'quorum.expected_votes must be configured'
808 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
810 When corosync starts to fail and you get the following message in the system log:
814 corosync[1647]: [QUORUM] Quorum provider: corosync_votequorum failed to initialize.
815 corosync[1647]: [SERV ] Service engine 'corosync_quorum' failed to load for reason
816 'configuration error: nodelist or quorum.expected_votes must be configured!'
820 It means that the hostname you set for corosync 'ringX_addr' in the
821 configuration could not be resolved.
824 Write Configuration When Not Quorate
825 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
827 If you need to change '/etc/pve/corosync.conf' on an node with no quorum, and you
828 know what you do, use:
834 This sets the expected vote count to 1 and makes the cluster quorate. You can
835 now fix your configuration, or revert it back to the last working backup.
837 This is not enough if corosync cannot start anymore. Here its best to edit the
838 local copy of the corosync configuration in '/etc/corosync/corosync.conf' so
839 that corosync can start again. Ensure that on all nodes this configuration has
840 the same content to avoid split brains. If you are not sure what went wrong
841 it's best to ask the Proxmox Community to help you.
844 [[corosync-conf-glossary]]
845 Corosync Configuration Glossary
846 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
849 This names the different ring addresses for the corosync totem rings used for
850 the cluster communication.
853 Defines to which interface the ring should bind to. It may be any address of
854 the subnet configured on the interface we want to use. In general its the
855 recommended to just use an address a node uses on this interface.
858 Specifies the mode of the redundant ring protocol and may be passive, active or
859 none. Note that use of active is highly experimental and not official
860 supported. Passive is the preferred mode, it may double the cluster
861 communication throughput and increases availability.
867 It is obvious that a cluster is not quorate when all nodes are
868 offline. This is a common case after a power failure.
870 NOTE: It is always a good idea to use an uninterruptible power supply
871 (``UPS'', also called ``battery backup'') to avoid this state, especially if
874 On node startup, the `pve-guests` service is started and waits for
875 quorum. Once quorate, it starts all guests which have the `onboot`
878 When you turn on nodes, or when power comes back after power failure,
879 it is likely that some nodes boots faster than others. Please keep in
880 mind that guest startup is delayed until you reach quorum.
886 Migrating virtual guests to other nodes is a useful feature in a
887 cluster. There are settings to control the behavior of such
888 migrations. This can be done via the configuration file
889 `datacenter.cfg` or for a specific migration via API or command line
892 It makes a difference if a Guest is online or offline, or if it has
893 local resources (like a local disk).
895 For Details about Virtual Machine Migration see the
896 xref:qm_migration[QEMU/KVM Migration Chapter]
898 For Details about Container Migration see the
899 xref:pct_migration[Container Migration Chapter]
904 The migration type defines if the migration data should be sent over an
905 encrypted (`secure`) channel or an unencrypted (`insecure`) one.
906 Setting the migration type to insecure means that the RAM content of a
907 virtual guest gets also transferred unencrypted, which can lead to
908 information disclosure of critical data from inside the guest (for
909 example passwords or encryption keys).
911 Therefore, we strongly recommend using the secure channel if you do
912 not have full control over the network and can not guarantee that no
913 one is eavesdropping to it.
915 NOTE: Storage migration does not follow this setting. Currently, it
916 always sends the storage content over a secure channel.
918 Encryption requires a lot of computing power, so this setting is often
919 changed to "unsafe" to achieve better performance. The impact on
920 modern systems is lower because they implement AES encryption in
921 hardware. The performance impact is particularly evident in fast
922 networks where you can transfer 10 Gbps or more.
928 By default, {pve} uses the network in which cluster communication
929 takes place to send the migration traffic. This is not optimal because
930 sensitive cluster traffic can be disrupted and this network may not
931 have the best bandwidth available on the node.
933 Setting the migration network parameter allows the use of a dedicated
934 network for the entire migration traffic. In addition to the memory,
935 this also affects the storage traffic for offline migrations.
937 The migration network is set as a network in the CIDR notation. This
938 has the advantage that you do not have to set individual IP addresses
939 for each node. {pve} can determine the real address on the
940 destination node from the network specified in the CIDR form. To
941 enable this, the network must be specified so that each node has one,
942 but only one IP in the respective network.
948 We assume that we have a three-node setup with three separate
949 networks. One for public communication with the Internet, one for
950 cluster communication and a very fast one, which we want to use as a
951 dedicated network for migration.
953 A network configuration for such a setup might look as follows:
956 iface eno1 inet manual
960 iface vmbr0 inet static
962 netmask 255.255.250.0
970 iface eno2 inet static
972 netmask 255.255.255.0
976 iface eno3 inet static
978 netmask 255.255.255.0
981 Here, we will use the network 10.1.2.0/24 as a migration network. For
982 a single migration, you can do this using the `migration_network`
983 parameter of the command line tool:
986 # qm migrate 106 tre --online --migration_network 10.1.2.0/24
989 To configure this as the default network for all migrations in the
990 cluster, set the `migration` property of the `/etc/pve/datacenter.cfg`
994 # use dedicated migration network
995 migration: secure,network=10.1.2.0/24
998 NOTE: The migration type must always be set when the migration network
999 gets set in `/etc/pve/datacenter.cfg`.
1003 include::pve-copyright.adoc[]